Key questions: - the key words used (and how they evolved over time) - publishing country (or first auttors country) - -articles per year/ the publishing year - (articles on marine SES) - maybe the document type
note I was not able to produce a map of affiliation countries. Unfortunately this was not possible because the ‘Affiliation’ metadata was not exported in the .ris file. See available columns below:
## [1] "Article ID"
## [2] "Article URL"
## [3] "Status"
## [4] "User Count"
## [5] "Users"
## [6] "Include"
## [7] "Population:"
## [8] "Intervention/Exposure"
## [9] "Outcome 1"
## [10] "Outcome 2 (not an exlcusion criterion!)"
## [11] "Unsure"
## [12] "User Notes"
## [13] "Title"
## [14] "Journal"
## [15] "Authors"
## [16] "Filenames"
## [17] "label"
## [18] "type"
## [19] "author"
## [20] "journal"
## [21] "abstract"
## [22] "year"
## [23] "doi"
## [24] "volume"
## [25] "issue"
## [26] "pages"
## [27] "language"
## [28] "issn"
## [29] "url"
## [30] "keywords"
## [31] "publisher"
## [32] "accession"
## [33] "CONC"
## [34] "FIND"
## [35] "METH"
## [36] "OBJE"
## [37] "RESU"
## [38] "SIGN"
## [39] "DA"
## [40] "DB"
## [41] "j2"
## [42] "n1"
## [43] "ST"
## [44] "a2"
## [45] "a3"
## [46] "stringDist"
*Explanation of topic models from revtools documentation**
Uses: Helpful for gaining an a priori impression of the dominant themes within search results. Limitations: Cannot define the topics a priori (although the number of topics sought can be specified) – groupings are a function of the data.
Plot of the ordination of the topic classifications from the topic model. Each point represents one article, with points colored according to the highest-weighted topic for that article. When you hover your mouse over a point, you can see the title of the article, and the topic classification.
The bar chart shows the number of articles within each topic. Hovering the mouse over a bar will show the keywords that define each topic.
You can see the difficulty with the topic model approach – is there really a meaningful difference in the concepts embodied by the different topics? I can’t really see anything but maybe to a more trained eye this may mean something.
There are many ways you could try to identify articles relevant for marine SES. Here I just experimented with a keyword matching search
First I assembled a list of keywords grouped by different descriptive factors we would like to extract information about. We then scanned the article title, abstract and keywords and counted the occurrences of matches to these terms (boolean response of “yes” for at least one match/article).
| Group | Group.name | Term | Term.type | |
|---|---|---|---|---|
| 1 | marine | marine | ocean | single |
| 2 | marine | marine | marine | single |
| 3 | marine | marine | sea | single |
| 4 | marine | marine | coast | single |
| 5 | marine | marine | shore | single |
| 8 | SES | SES | social ecological system | expression |
| 9 | SES | SES | social ecological systems | single |
We then constructed a document feature matrix of all the terms (a type of feature) * the articles where occurrences were found. To calculate when terms occurred together in the same document, we constructed a feature co-occurrence matrix with dimensions V*V where V=the number of terms (features)
| social_ecological_system | social_ecological_systems | |
|---|---|---|
| ocean | 2 | 0 |
| marine | 2 | 0 |
| sea | 20 | 0 |
| coast | 3 | 0 |
| shore | 0 | 0 |